Skip to content

Latest commit

 

History

History
60 lines (44 loc) · 2.81 KB

treemap.md

File metadata and controls

60 lines (44 loc) · 2.81 KB
Raw

cldfviz.treemap

Plot values for a parameter of a CLDF dataset on a map linked to a tree. The core functionality is implemented in Florian Matter's lingtreemaps package and all its configuration options are available from cldfviz.treemap as options prefixed with --ltm-.

cldfviz.treemap adds the ability to run lingtreemaps with data from CLDF datasets, exploiting additional data from LanguageTables to allow for configurable matching between data and tree.

As an example, we plot values of WALS feature 88A for languages in a couple of Indo-European genera against the Glottolog classification for Indo-European.

$ cldfbench cldfviz.treemap wals-2020.3/ 88A --tree indo1319 \
--output tm.svg --open --glottolog-cldf glottolog-cldf-4.7/ \
--language-filters '{"Genus":"Germanic|Romance|Celtic"}' --ltm-text-x-offset 0.5

Tree specification

The tree can be specified in various ways:

  1. Using the --tree option which accepts
    • a Glottocode - in which case the Glottolog sub-classification tree for the languoid specified by --tree is used (see above),
    • a Newick-formatted string,
    • a path name to an existing file containing the Newick-formatted tree.
  2. Using the --tree-dataset (and --tree-id) options to select a tree provided in the TreeTable of a CLDF dataset.

Since the WALS CLDF data contains the (shallow) trees of its Genealogical Language List, we can plot the same data again as follows:

$ cldfbench cldfviz.treemap wals-2020.3/ 88A --output tm.svg --open \
--language-filters '{"Genus":"Germanic|Romance|Celtic"}' --ltm-text-x-offset 0.5 \
--tree-dataset wals-2020.3/ --tree-id family-indoeuropean \
--tree-label-property ID

Since WALS contains coordinates for all its languages, we do not have to access Glottolog data anymore.

Matching languages in data and tree

Linking the languages in the parameter's dataset to languages in the tree can be configured via the --glottocodes-as-tree-labels and --tree-label-property options. The former will re-name tree labels to Glottocodes using the mapping given in the --tree-dataset's LanguageTable. The latter specifies a column in the parameter's LanguageTable values of which are used to match languages to tree node labels.

In the second examples above, we used WALS family trees, which use the 'ID' values of the LanguageTable as node labels. Thus, we had to specify --tree-label-property ID to make sure datapoints can be matched with nodes in the tree.